parse_date_time fails for abbreviated month names in locales where abbreviation includes "."

[EmilBode]

I suspect this may be a regression bug due to the fix for issue #781

When trying to parse dates with parse_date_time, I noticed unexected NAs when I used abbreviated month names in French and Spanish

Reprex

library(lubridate)
Sys.setlocale(locale='Spanish_Spain')
# Yes, locale specification is superfluous, just to be sure it's using the right one
lubridate::parse_date_time('02 ene. 2015 12:48', '%d %b %Y %H:%M', locale='Spanish_Spain')
# > [1] NA
# > Warning message:
# > 1 failed to parse. 
# Same result without dot
lubridate::parse_date_time('02 ene 2015 12:48', '%d %b %Y %H:%M', locale='Spanish_Spain')
# > [1] NA
# > Warning message:
# >  1 failed to parse. 
# Does work with strptime (and dot)
strptime('02 ene. 2015 12:48', '%d %b %Y %H:%M')
# > [1] "2015-01-02 12:48:00 CET"

Sys.setlocale(locale='French_France')
lubridate::parse_date_time('02 janv 2015 12:48', '%d %b %Y %H:%M', locale='French_France')
# > [1] NA
# > Warning message:
# >  1 failed to parse. 
# And same with a dot
# But note that this does work, probably because "mars" is not an abbreviation, thus always without dot:
lubridate::parse_date_time('02 mars 2015 12:48', '%d %b %Y %H:%M', locale='French_France')
# > [1] "2015-03-02 12:48:00 UTC"

Environment

• R 3.6.2 under Rstudio 1.2.5033
• lubridate 1.7.8
• Other loaded packages: just the basics: base, methods, dataset, utils, grDevices, graphics, stats
• Windows 10, 64 bit, Dutch version
• OS locale "LC_COLLATE=Dutch_Netherlands.1252;LC_CTYPE=Dutch_Netherlands.1252;LC_MONETARY=Dutch_Netherlands.1252;LC_NUMERIC=C;LC_TIME=Dutch_Netherlands.1252" (as specified by Sys.setLocale())

Diagnostics output of month names output according to strptime/format and lubridate:

Sys.setlocale(locale='French_France')
# > [1] "LC_COLLATE=French_France.1252;LC_CTYPE=French_France.1252;LC_MONETARY=French_France.1252;LC_NUMERIC=C;LC_TIME=French_France.1252"
format(seq.Date(as.Date('2015-01-01'), by='month', length.out = 12), '%b')
# >  [1] "janv." "févr." "mars"  "avr."  "mai"   "juin"  "juil." "août"  "sept." "oct."  "nov."  "déc." 
month(seq.Date(as.Date('2015-01-01'), by='month', length.out=12),label = TRUE)
# >  [1] janv févr mars avr  mai  juin juil août sept oct  nov  déc 
# > Levels: janv < févr < mars < avr < mai < juin < juil < août < sept < oct < nov < déc
Sys.setlocale(locale='Spanish_Spain')
# > [1] "LC_COLLATE=Spanish_Spain.1252;LC_CTYPE=Spanish_Spain.1252;LC_MONETARY=Spanish_Spain.1252;LC_NUMERIC=C;LC_TIME=Spanish_Spain.1252"
format(seq.Date(as.Date('2015-01-01'), by='month', length.out = 12), '%b')
# >  [1] "ene." "feb." "mar." "abr." "may." "jun." "jul." "ago." "sep." "oct." "nov." "dic."
month(seq.Date(as.Date('2015-01-01'), by='month', length.out=12),label = TRUE)
# >  [1] ene feb mar abr may jun jul ago sep oct nov dic
# > Levels: ene < feb < mar < abr < may < jun < jul < ago < sep < oct < nov < dic

[vspinu]

Thanks for the report. I have added a test for non-english locale to avoid such embarrassing bugs in the future.